Eyetohand

Eyetohand refers to the coordination of eye movements and hand actions to interact with the environment. It underpins visually guided manipulation, enabling tasks such as reaching, grasping, and avoiding obstacles based on what is seen.

In humans, eyetohand relies on visuo-motor integration. The dorsal visual stream processes spatial information about objects and their location, while the parietal and premotor regions translate this information into hand trajectories. Gaze often precedes hand movement, and rapid eye movements (saccades) help sample scenes that guide action. Visual feedback from the scene, together with proprioceptive and motor signals, supports online corrections during reaching.

In robotics and human-computer interaction, the term is used to describe systems in which an external camera (eye) provides visual input to guide a robotic hand or a user interface, as opposed to an eye-in-hand setup in which the camera is mounted on the manipulator. Eye-to-hand configurations offer a stable world-centric view and are common in industrial manipulation, teleoperation, and assistive devices, though they introduce calibration, latency, and field-of-view trade-offs.

Research in eyetohand covers sensor fusion, gaze-based intention estimation, visual servoing, and predictive control to reduce lag between perception and action. Challenges include accurate gaze-to-action mapping, depth perception, occlusion, and dynamic scene changes. Ongoing work aims to improve robustness in unstructured environments and to extend human-robot collaboration.

The term is sometimes written eyeto-hand or eye-to-hand; it is closely related to general eye-hand coordination and visuomotor integration.